Migration from DS 7.5.1A Server Edition to DS 8
Moderators: chulett, rschirm, roy
Raftsman wrote:-
We have imported our 7.5.x2 jobs to 8.0.1 and been running the two versions alongside each other. We have also raised a number of tickets over the last couple of weeks and are awaiting responses:
1) Parameter sets don't work with environment variables and values over-ridden at run time do not get passed from a seuqence to lower level jobs and sequences.
2) Importing some jobs gives different behaviour due to a change in the way operators are combined. Functionally there appears to be no effect, but the job stops pipelining data. At least it's easy to spot these cases.
3) Sorts won't allow more than about 550MB memory per node. This was also a problem in 7.5x2.
4) Sequential file stage doesn't always take note of the property indicating that the first line is column names (header).
What sort of problems have you been having and have the tickets been fixed or had workarounds produced by IBM?We are coming up on three months now and we still do not have our Windows version of 8.0.1 working. IBM has been contacted and we are continually opening tickets. I would be very interested to hear if anyone using the Windows 8.0.1 version has it working properly. We use an 8 node box on a P1 platform.
We have imported our 7.5.x2 jobs to 8.0.1 and been running the two versions alongside each other. We have also raised a number of tickets over the last couple of weeks and are awaiting responses:
1) Parameter sets don't work with environment variables and values over-ridden at run time do not get passed from a seuqence to lower level jobs and sequences.
2) Importing some jobs gives different behaviour due to a change in the way operators are combined. Functionally there appears to be no effect, but the job stops pipelining data. At least it's easy to spot these cases.
3) Sorts won't allow more than about 550MB memory per node. This was also a problem in 7.5x2.
4) Sequential file stage doesn't always take note of the property indicating that the first line is column names (header).